@altruios @jadedtwin @WhiteCatTamer @Eatsbluecrayon

Sounds like determinism.

> From the count of neurons firing, to their relations and positions: every “emotion” can be described with numbers.

As I remember, we can't do that because:

1) we still don't know how the human brain and consciousness works

2) On MRI we can see some common (for each human being) zones in brain, firing up when some common emotion occurs, thats all.

3) https://aeon.co/essays/your-brain-does-not-process-information-and-it-is-not-a-computer

@Eatsbluecrayon @jadedtwin @WhiteCatTamer @altruios

One would have to define what ”expressed in numbers” and ”described with numbers” mean. Is Pi a number? And if it is, does the symbol express it, or is it an expression for the inability to describe that function as a number?

But the main thing is this: Language has no fixed meaning. A symbol (like a word) is not one thing, but many. And it shifts.

Math doesn’t.

@altruios @jadedtwin @WhiteCatTamer @Eatsbluecrayon
I see this the other way around.

Humans use numbers to understand things. Every new thing is assigned measurements and studied. This helps a lot in understanding how everything works.

When studying something new, we start by assigning them and seeing how they behave. We need different rules for different processes.

The numbers are not the starting point, they are the way we try to make sense of things that are very complex. If we start with only numbers, we will not be able to figure out anything. We need to give the numbers meaning, and discover the rules behind them.

I'm reminded of Socrates. If you keep asking why, there will be a moment where you don't know. Why do things fall? Well, because of gravity. Why does gravity exist? Well, the earth's mass attracts more mass. This can be explained by diving into molecules and atoms and ions and neutrons and whatnot, but why are there atoms? And why was the big bang? And why does anything exist at all?

For us to assign numbers, we need to understand so much more. And a computer is bound by the logic we give it. Yes, it can make some things up by itself, but those things can be just as wrong as our age-old belief of the flatness of the earth. And the computer will lack knowledge we didn't know it needed. Or things we don't know ourselves.

The world is a magical place, and we are by no means done discovering and making sense of it. ❤️

@altruios @jadedtwin @WhiteCatTamer @Eatsbluecrayon Perhaps you've not read, or understood, Godel's Theorems etc.

Maths *is* amazing, for example, in being "unreasonably good" at describing the workings of the universe to incredible precision.

HOWEVER, as well as Godel, we also have hand-wavy math of QFT e.g. the "Yang–Mills existence and mass gap" Millennium Prize Problem.

Let alone struggles to describe "complex", "chaotic" etc. systems.

Maths limitations are as incredible as what it does.

@worik @jadedtwin @WhiteCatTamer @Eatsbluecrayon

LLMs are not "fancy autocorrect" but they ARE fancy autocomplete. Most of their training is just any kind of text that makes sense, and only on the final stages of training you feed it conversations between "user" and "assistant", to specialize it in following instructions. The illusion of a personality disappears as soon as you're allowed to request the LLM to keep completing the text: after autocompleting the answer of the "assistant" it starts autocompleting the message of the "user". LLMs are impersonation machines, it's just that in most cases they're made to impersonate the "assistant" side. When I have access to the raw text autocompletion of a LLM I have a bit of fun seeing what it autocompletes as the user, or how it autocompletes a conversation that is not a user-assistant conversation. For me it demystifies this magic thinking that people have about LLMs.

The "thinking" that some LLMs do is just an extension of a technique called "chain of thought", to make it have more information in the context and to be able to resolve contradictions. It doesn't need to be a "thought" in the same language or even in natural language at all, it works just as well if it appears as random symbols to us. It's just that deepseek trained it to be readable. It's not real thinking. It works better than without it, certainly, but real thinking would involve much more than just generating a bunch of stuff to improve the quality of next generations. Actual thinking involves abstract non-verbal thought as well as being able to learn from experience (even from just one single experience).

The only way LLMs learn nowadays is by "training" on a lot of data so it eventually recognizes patterns, and you can't just train it on one conversation to make it learn, it doesn't work like that.